The software captures the patient’s voice and processes the sounds to be matched with the software’s dictionary for interpretation of the pronunciation.
Hsieh T, Funamura JL, Roth C, Rubin Z, Kurniawan S, Tollefson TT. Developing a Novel Speech Intervention iPad Game for Children With Cleft PalateA Pilot Study. JAMA Facial Plast Surg. 2015;17(4):309-311. doi:10.1001/jamafacial.2015.0164
Children with cleft palate (CP) require multidisciplinary care, including palatoplasty and speech therapy (ST). There is currently an unmet need for ST for children with CP in resource-poor and rural areas owing to financial burden and shortages of transportation and health care professionals.1 Poor speech in these children, compounded by the aesthetic stigmata of cleft lip and/or palate, can have social and economic consequences in adulthood.
Speech therapy is needed in up to 68% of children with CP, often for compensatory articulation errors such as the glottal stop, where the vocal folds adduct as a substitution for high-pressure, oral plosives.2,3 Therapy includes establishment of target sounds, generalization of the new habits, and maintenance at home.3 Our study tested the feasibility and replay value of a game-based supplementary speech tool that is designed to augment office-based therapy by engaging parents and children in underserved areas.4,5
Ten children with CP and/or a cleft lip (aged 2-7 years) were identified as needing ST between May 29, 2013, and July 26, 2013. Children with cognitive impairment or profound hearing loss were excluded, as they needed to complete audio commands. The University of California, Davis, Institutional Review Board approved this study. Written consent was obtained from the children’s parents.
Software engineers, surgeons, and speech-language pathologists (SLPs) developed this interactive speech program for the iPad (Apple Inc). The game requires the child to produce targeted phrases to guide the main character through adventures by following on-screen prompts. The first speech commands were chosen to function with the speech recognition software (SRS) (OpenEars; Politepix UG) and to target the plosives /p,t,k/, which are erroneously spoken in glottal stops.3 The game is designed to provide carryover of the plosive skills, which were introduced in face-to-face ST.
In a series of stories, the game software on Cocos2D (Apportable Inc) produces the target phrase for the child to repeat. The SRS converts the child’s speech into discrete sounds. If the target words are pronounced incorrectly, they are highlighted on the screen and the child has another opportunity to pronounce them (Figure).
The SLP rated the child’s speech using the standardized speech assessment protocol before initiating the game. The time taken for the child to complete the game (attention span, 5-7 minutes)6 while observed by the SLP, surgeon, and parents was recorded. During game play, the SLP assessed the speech pronunciations and compared the pronunciations with the assessment of pronunciations provided by the game’s SRS. The questionnaire for parental feedback (5-item Likert-type scale) was evaluated with a Kruskal-Wallis test (R, version 2.10.0).
Ten children were enrolled in the study (Table). Mean (SD) game completion time was 164.3 (50.5) seconds. Every child completed the game storyboards with the designed repetition. Speech recognition was correct on the first attempt in 3 of 5 target phrases (“Put on boots,” “Wear a hat,” and “Cross the bridge”) and on the third attempt for the other 2 phrases (“Pop a balloon” and “Open the door”). There was high concordance between the game and real-time SLP scoring of responses.
Parental responses regarding the child’s engagement, replay value of the game, appropriate difficulty (all but one parent), positive rewards, and clear goal of the game produced mean scores higher than 4. Mean scores were not significantly different among the parental responses (P = .30).
This pilot study was designed to test the feasibility of a novel iPad game software as a supplemental ST tool that may improve generalization of accurate speech production in children with CP. The game completion time was reasonable and within the child’s attention span. Participant responses were effectively captured by the game’s SRS with high accuracy. The children correctly made speech adjustments to trigger the game to move ahead to the next screen. Parents perceived the game as engaging, with high replay value, and noted better understanding of the ST strategies.
Future game improvements include an improved SRS, new game storyboards that are stratified for age, and Internet-based, real-time feedback for the SLP and child. As many of us experience with smartphones’ voice recognition weaknesses, some sounds do not match the system’s phonetic dictionary but are accurate enough to be understood by an observer. A blinded SLP will score the child’s videotaped speech samples from the gameplay to enable the system programming to narrow this gap.
Corresponding Author: Travis T. Tollefson, MD, MPH, Department of Otolaryngology–Head and Neck Surgery, University of California Davis Medical Center, 2521 Stockton Blvd, Ste 7200, Sacramento, CA 95817 (email@example.com).
Published Online: April 16, 2015. doi:10.1001/jamafacial.2015.0164.
Author Contributions: Dr Tollefson had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hsieh, Funamura, Rubin, Kurniawan, Tollefson.
Acquisition, analysis, or interpretation of data: Hsieh, Funamura, Roth, Rubin, Tollefson.
Drafting of the manuscript: Hsieh, Rubin, Kurniawan, Tollefson.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Hsieh, Rubin, Tollefson.
Obtained funding: Hsieh, Funamura, Rubin, Tollefson.
Administrative, technical, or material support: Hsieh, Roth, Rubin, Kurniawan, Tollefson.
Study supervision: Hsieh, Roth, Tollefson.
Conflict of Interest Disclosures: None reported.